Random Number Generator (RNG) - Not so random

I’ll start by saying that overall I believe when taking in large amounts of data with hero rolls, ascension rolls, etc that things average out the way that they’re “supposed” to (albeit with really slim odds for gaining certain heroes/items).

What I have noticed though consistently and I know many others have is that there is a “clumpiness” to the randomness. Successful hero rolls seem to happen in clumps. Unsuccessful ones seem to as well. It seems that you can be just as likely to get several great heroes in a 10x roll as to get just one. They seem to come together in groups like this in a manner which is not so “random”.

This then seems to get offset by large amounts of 10x rolls that yield nothing of great value.

This is great if you get one of the amazing 10x rolls, but can be complete rubbish if you’re on the end where you’re getting the losing rolls that must exist to offset the clumped together amazing rolls.

I guess like I’ve seen Rook say “randomness” here pretty much is another way of saying luck, but I think more consistency and less “clumpiness” would be better.

Cheers,

DB

2 Likes

Obviously,DB, you don’t shoot craps.

2 Likes

Yes, i always had the same impression when it comes to summons and (lately) even map loot.
All the good stuff come togheter, as if there was a specific time to get them.

The only thing i’m missing, is how it can be this way on the first day of the month.

I mean, i’m quite sure people assault the gate as soon as the new HotM is avalaible, but results for the players (even only with some minutes of delay) are very different.

So at least i guess on the first day it isn’t like that.

2 Likes

I get the feeling that the board is not random at all. I notice this in war and titan mostly. On titan battels i go with 4 or 5 strong colour heroes and sometimes i can’t fill the ss of very fast hero in 6 moves or even more. In war also. Mostly when i go with 3-2 colour team againts the oponent, no colour on the board.
This has become so obvious that it is frustrating

1 Like

Here is a reference if we have bad board.

2 Likes

I get this, but still tends to be impossible to save the bad board. Still stand by my statement. Can’t fill.a very fast hero, and still while salvageing the bad board I only fill oponents mana

1 Like

I’ve been using mono teams for war for a while now. It seems I generally get 1 good board and 2 average boards for every bad board. Last war I did have 2 really bad boards but I also had 2 really good ones where I could take out 3900 teams pretty quickly with a 3200 and 3400 team.

2 Likes

@ffphier happens to me sometimes. But in general bad luck is with me, and so i just want to emphasise my opinion that the boards are not random

1 Like

I’m a developer, and I have worked with randomizers, hashing and seeds.

I will tell you one thing, this is not very random. It is actually pretty interesting to know how they manage to create such a poor randomizer (I’m not whining here, just giving my fair opinion). Given the amount of available heroes, the default randomizer in all the languages I’ve used, couldn’t even produce so many duplicates unless you use a very stupid seed.

Fascinatingly poor, from a developer perspective.

9 Likes

As a fellow developer, I have to disagree. They give you the percentages. Plug them into a simulation, hook up all the prngs you want, run some tests.

Remember you’re one small slice of everyone being served by their server side RNG.

1 Like

I don’t understand this answer. The OP is talking about duplicate heroes. There are no percentages at all for specific heroes. Only for the chances of get one of certain groups of heroes like 3*, 4*, 5* for example. How is that related to specific heroes?

I don’t know who you’re talking to, but that’s not what the OP is talking about. Drubas (not the OP) is talking about dupes, but dupes are easily explained. The pool of heroes is very small for any given summon, and as you correctly point out, the percentages are banded by rarity. Most of your draws are going to be 3 stars, point blank. Since you have 3 or so 3 stars that are unique to each pool, and the rest are s1 heroes, you are more likely to draw a hero you already have.
That’s just how it is, even assuming you have a flat chance per hero within those bands, which, as simulation testing shows, you likely do. The longer you play, the more dupes you get.

This even happens for legendary pulls - for instance I have had all s1 5 stars for several months now, except 3. The missing heroes are just less likely to be filled in because that’s how odds work. I have a 1/whatever chance to draw a 5 star, then I have a 1/25(?)ish chance to get a specific hero, each draw. The odds of me finally getting those last two are low, each draw. Eventually it will happen, but in the meantime, dupes.

Sorry, mix up at my side. I was filling in the lost minutes in between looking here during work and Drupas’ post happened to be on top of my screen when I opened it again. So I assumed that he was the OP and that you replied to him, especially since you started by staying that you’re a fellow developer.

Odd go against your intuition, this is true. Put 23 people in a room and the odds that two of them have the same birthday is about 50%. You just don’t expect that. Same thing happens here and it’s enlarged by our human tendency to see patterns and make connections everywhere.

Wow, I have by far the most 5* heroes in red and I’m getting fewer hidden daggers than anything else. Look at that, I have five 4* troops for every colour, but only one for blue. Can you believe it, not a Single HOTM in one and a half years and then two within 5 minutes. And yes, also: I have only 18 5* heroes and five of them are Elena. These have all been true in my case btw. Odds are weird.

Whats a SEED or hashing? I’m just curious

Seed is the input to the RNG. Usually it’s just an integer. https://en.m.wikipedia.org/wiki/Random_seed

Hashing is taking the data and running it through mathematical changes to map something of arbitrary size (like any integer) to something in a specific range (0-100). https://en.m.wikipedia.org/wiki/Hash_function

2 Likes

Anyone who claims there was is no tinkering please explain one month ago everyone get 5th grade chests from goblin ballon in same week. That is also powered by RNG :slight_smile:

Also don’t forget that items that they can assign specific drop ratio to specific items on RNG. Those variables can easily be tinkered without any update on client side if needed.

1 Like

“Clumpiness” is to be expected in random sequences. There are many (like really, really a lot) more distributions which are clumpy than distributions which are smooth.

2 Likes

Odds are not disclosed for goblin chest drop, they could have an entirely different system for generation. This is not the slam dunk argument you think it is. While there was likely some manipulation, it wasn’t, for whatever reason, universal. Some people got it at the start of a chest, some people got it second, and some never saw it (though all this is anecdotal).

And no, none of that is handled client side. All RNG is server side, for good reason.

1 Like

All I know is that, nobody here can consistently predict accurately in sequence which heroes or troops are to appear when you do single pulls or a x10 or a x30 pulls, or which tiles appear in what position or order or number when you match tiles and whether such move will trigger 2, 3, 4, 5, 6, 7, or an x number of cascades, or the exact loot to be obtained from various chests or rewards. When one can do that on a consistent basis, only then will I believe that there is really no randomness in the game since one can already predict the outcome. Predicting something successfully is totally the polar opposite of randomness, IMHO.

2 Likes

Which is like saying you can’t infer if a dice is rigged without knowing what all the other dices in the world are doing :relieved:

1 Like

Cookie Settings